Emergence of transmissible SARS-CoV-2 variants with decreased sensitivity to antivirals in immunocompromised patients with persistent infections

Abstract

We investigated the impact of antiviral treatment on the emergence of SARS-CoV-2 resistance during persistent infections in immunocompromised patients (n = 15). All patients received remdesivir and some also received nirmatrelvir-ritonavir (n = 3) or therapeutic monoclonal antibodies (n = 4). Sequence analysis showed that nine patients carried viruses with mutations in the nsp12 (RNA dependent RNA polymerase), while four had viruses with nsp5 (3C protease) mutations. Infectious SARS-CoV-2 with a double mutation in nsp5 (T169I) and nsp12 (V792I) was recovered from respiratory secretions 77 days after initial COVID-19 diagnosis from a patient sequentially treated with nirmatrelvir-ritonavir and remdesivir. In vitro characterization confirmed its decreased sensitivity to remdesivir and nirmatrelvir, which was overcome by combined antiviral treatment. Studies in golden Syrian hamsters demonstrated efficient transmission to contact animals. This study documents the isolation of SARS-CoV-2 carrying resistance mutations to both nirmatrelvir and remdesivir from a patient and demonstrates its transmissibility in vivo.

Fig. 1. Emergence of SARS-CoV-2 variants in immunocompromised patients.

a The treatment course and sample collection from 15 patients with persistent SARS-CoV-2 infections. Clade and PANGO lineage designations are included with Patient IDs. All time points along the x axis are referenced from the date of each patient’s initial positive COVID-19 test result, with day 0 marking the date of the first positive test at the New York Presbyterian (NYP) Hospital (dashed gray line). Treatment courses are specified along the y axis for each patient and colored based on the treatment category. Nasopharyngeal swab samples successfully sequenced and used for analyses are represented as black points along the solid black line for each patient. The total number of unique single-nucleotide variants (SNVs) found ≥2% across the genome (top) over the course of the infection in patients b 11595, c 16902, and d 17072 compared to their treatment courses with nirmatrelvir-ritonavir (pink) or remdesivir (tan). SNVs are determined by comparing each sample to their respective clade consensus sequences (“Methods”). The relative frequency of non-synonymous SARS-CoV-2 SNVs in nsp5 and nsp12 coding regions during infection are shown below. Variant data are grouped by the coding region, amino acid position, and amino acid residue, with the color representing the amino acid and the shape indicating whether the variant was found above (passes, circle) or below (fails, triangle) our detection cutoffs (“Methods”). All time points along the x axis are referenced from the date of each patient’s initial positive COVID-19 test result, with day 0 marking the date of the first positive test at the NYP hospital.

Fig. 2. Sensitivity of SARS-CoV-2-nsp5^T169Insp12^V792I virus to nirmatrelvir and remdesivir in vitro.

a Experimental layout showing the treatment plans to study the sensitivity of the SARS-CoV-2-nsp5^T169Insp12^V792I virus to antiviral therapies. b Nirmatrelvir resistance of SARS-CoV-2. Vero E6 cells were treated with increasing doses of nirmatrelvir (0.05–25 µM) following three treatment conditions (pre-, simultaneous, and post-treatment) and infected with 200 TCID₅₀/well of SARS-CoV-2-nsp5^T169Insp12^V792I and WT viruses. Virus titers in supernatant of treated/infected cells was quantified by limiting dilution and the percent inhibition value was calculated to obtain the 50% inhibitory concentration (IC₅₀). Results represent the average of 4 (pre- and simultaneous treatment) and 6 (post-treatment) replicates and three independent experiments. c Remdesivir resistance of SARS-CoV-2. Vero E6 cells were treated with increasing doses (0.09–100 µM) of remdesivir following three treatment regimens and infected with 200 TCID₅₀/well of SARS-CoV-2-nsp5^T169Insp12^V792I and WT isolates. Virus titers in supernatants were quantified by limiting dilution and the percent inhibition value was calculated to obtain the 50% inhibitory concentration (IC₅₀). Results represent the average of four replicates from two independent experiments. a Created with BioRender.com released under a Creative Commons Attribution-NonCommercial-NoDerivs 4.0 International license.

Fig. 3. Efficacy of combination nirmatrelvir and remdesivir therapy against SARS-CoV-2-nsp5^T169Insp12^V792I virus in vitro.

Vero E6 cells were treated with indicated concentration of nirmatrelvir or remdesivir alone or in combination at pre-infection (a), at the time of infection (b) or at post-infection (c) and infected with 200 TCID₅₀/well of SARS-CoV-2-nsp5^T169Insp12^V792I. After 48 h, the cell monolayer was fixed and stained with SARS-CoV-2 N specific monoclonal antibody (red) and the nucleus was counterstained with DAPI (blue). White Bar 1 mm. d Virus titration (TCID₅₀.mL⁻¹) in the supernatant of nirmatrelvir, remdesivir or combination treatment from (a–c). The results represent mean ± SEM of four replicates from two independent experiments. Two-way ANOVA with Bonferroni multiple comparison test, *P < 0.05 and ***P < 0.001.

Fig. 4. Structural analysis of nsp5 T169I and nsp12 V792I substitutions.

Comparison of nsp5 (a) and nsp12 (b) protein models between WT (NYI45-21) and mutant nsp5^T169I and nsp12^V792I proteins. a, b, left panel: WT: Surface representation of nsp5 (magenta) and nsp12 (cyan) proteins with specific amino acid substitutions highlighted in the inset for each protein. The position of substituted amino acids is depicted in yellow sticks with proper labelling. a, b, middle panel: Superimposed Image: Cartoon representation of the superimposed image of WT (magenta/cyan) and nsp5^T169I and nsp12^V792I proteins (green/red), depicting specific substitutions in the inset box. a, b, right panel: nsp5^T169I and nsp12^V792I: nsp5 (magenta) and nsp12 (cyan) proteins with specific amino acid substitutions highlighted in inset for each protein in box. The position of the substituted amino acids is shown in yellow sticks with proper labelling. Note: The protein models of the nsp5^T169I and nsp12^V792I proteins was generated using Swiss MODELLER, resulting in RMSD values of 0.003 Å for nsp5 and 0.026 Å for nsp12 proteins, indicating high structural similarity. Surface representation of nsp5 proteins of (c) WT and (d) nsp5^T169I protein docked to nirmatrelvir drug molecules shown in the inset using PyMOL analysis tool. 2D depiction of nsp5 protein residues interactions with nirmatrelvir drug molecules via LIGPLOT⁺ tool where hydrogen bonds are shown in green dotted lines with a distance in Ångström (Å) and hydrophobic interactions are shown in curvature. Surface representation of nsp12 proteins of e WT and f mutant nsp12^V792I docked remdesivir drug molecules shown in inset for each using PyMOL tool. 2D depiction of nsp12 protein residues interactions with remdesivir drug molecule via LIGPLOT⁺ tool where hydrogen bond are shown in green dotted lines with a distance in Å and hydrophobic interactions are shown in curvature. The colour scheme is consistent throughout the figures, with magenta representing nsp5 and cyan representing nsp12 of WT isolate, and green representing nsp5 and red representing nsp12 of nsp5^T169Insp12^V792I isolate with substituted amino acid highlighted in yellow. Note: BE Binding energy.

Fig. 5. The SARS-CoV-2-nsp5^T169Insp12^V792I virus efficiently transmitted to contact golden Syrian hamsters.

a Experimental design. b Changes in body weight of hamsters following intranasal inoculation of SARS-CoV-2-nsp5^T169Insp12^V792I and WT viruses and in contact animals throughout the 14-day experimental period. c SARS-CoV-2 RNA load in oropharyngeal swabs quantified by rRT-PCR. (b-c) Data represents mean ± SEM of four animals/group/timepoint. d, e Infectious SARS-CoV-2 loads in oropharyngeal swabs determined using endpoint dilutions and expressed as TCID₅₀.mL⁻¹. f SARS-CoV-2 RNA load in nasal turbinate, trachea and lungs quantified by rRT-PCR. g Neutralizing antibody responses to SARS-CoV-2 assessed by virus neutralization assay (100% neutralization) in serum. d–g Data represents median of four animals/group/timepoint. a Created with BioRender.com released under a Creative Commons Attribution-NonCommercial-NoDerivs 4.0 International license.